DESCRIPTION: Diet is likely to play an important role in maintaining health and in the prevention or promotion of disease. Epidemiologic studies have tried to clarify the links between diet and chronic diseases; while a number of nutritional factors have been linked to the incidence of cardiovascular disease and diabetes mellitus in observational studies, the search for dietary risk factors of cancer has been rather disappointing, despite plausible biological mechanisms. One possible explanation for the observed lack of association is measurement error. Another contributor may be inadequate analytic methods. The proposed project will utilize the most appropriate dataset available worldwide to address these questions. The data available from the Cambridge, UK, arm of the European Prospective Investigation of Cancer and Nutrition (EPIC) are unique in comprising 2 food frequency questionnaires (FFQs), 2 Seven-day diet records, 2 24-hour recalls, as well as blood and 24h-urine samples from 25,000 men and women. During 3 visits to Cambridge with a duration of 3 months each the following hypotheses will be investigated: 1. Correlated error within and between diet assessment instruments may distort the diet-disease relation and mislead in the attempt to identify the most important dietary predictor(s) of disease. 2. Understanding the structure of the error correlation provides the opportunity to differentiate between foods and nutrients that may be measured well with an FFQ and those which are not. 3. The analytical approach to dietary data should employ dietary patterns or food groups rather than individual foods or nutrients. Results will be compared to those obtained from the Nurses? Health Study.